Home | Legals | Sitemap | KIT

Autonomous Capacity and Workload Management in Virtualized Service Infrastructures (DFG funded)

Autonomous Capacity and Workload Management in Virtualized Service Infrastructures (DFG funded)

Deutsche Forschungsgemeinschaft (DFG)

IT service management (ITSM) comprises the planning, development, and maintenance of IT services to customers. Although many new industry standards like ISO 20000 have been developed, the field is lacking established methods for efficient IT service management in elastic infrastructures such as corporate clouds, i.e. for dynamic workload management, resource allocation and service management process optimization. We develop new decision models and methods to support tasks such as adaptive prioritization mechanisms, server consolidation, dynamic capacity and workload management, performance prediction, and dimensionality and complexity reduction in data centers.

Research Streams

Scalable Control Mechanisms for Virtualized Data Center Operations: We develop an autonomic control infrastructure for virtualized data centers, which predicts future workload for various business applications and aligns resource capacities provided for virtual servers according to pre-determined service level objectives in order to minimize total energy consumption.

Risk-Aware Transition Management in Service Networks: In this project we develop models for estimating the business impact of risk resulting from IT service changes. The model takes into account the network of dependencies between services and various probabilistc sources like stochastic change-related downtime. Furthermore we derive decision modelsfor scheduling changes with minimum business impact.

Complexity and Workload Management based on Geometric Algebra: We apply dimensionalilty reduction techniques to multivariate workload data in order to reduce the complexity of decision problems in data centers. By capturing workload features and interpreting them as indicators for workload profiles, uncertainties, and complementarities amongst workloads, we address tasks such as server consolidation, anomaly detection, or capacity re-allocation problems.